help: how to determine wide-sense stationarity from a graph?

Hi, everyone

please tell me how to determine that a random process is wide-sense
stationarity from it's autocorrelation graph or power spectrum graph.

Thanks

net

netbsd7@hotmail.com skrev:
> Hi, everyone
>
> please tell me how to determine that a random process is wide-sense
> stationarity from it's autocorrelation graph or power spectrum graph.

You can't.

Stationarity is a property of the process that generates the data,
that relates *all* data to each other. You have only a sample
available.

The "best" you can do is to determine whether the sample is
stationary or not, i.e. whether the statistics seem to change
during the frame you have available. If you somehow see that,
say, the variance of a noise sequence changes over the frame,
you can argue that the process is nonstationary.

Stationarity can be *assumed* in order to simplify the analysis
of the data, though, but never determined.

Rune